Laser market prediction information generation method and device

Abstract

The embodiment of the invention provides a laser market prediction information generation method and device. The method comprises the following steps: acquiring multi-dimensional market information corresponding to a laser in response to a received market prediction request for the laser; preprocessing the multi-dimensional market information corresponding to the laser, and classifying the preprocessed multi-dimensional market information according to a time dimension, a product type dimension and a region dimension to obtain multiple types of text information; performing feature extraction on the plurality of types of text information to obtain index information corresponding to the plurality of types of text information, and constructing knowledge graph information corresponding to the laser based on the index information corresponding to the plurality of types of text information; and calling the trained market prediction model according to the knowledge graph information corresponding to the laser, and generating market prediction information corresponding to the laser. The method improves the accuracy of market prediction information.

Description

Method and apparatus for generating laser market forecast information Technical Field

[0001] This application relates to the field of optical measurement technology, and in particular to a method and apparatus for generating laser market forecast information. Background Technology

[0002] Lasers, as core components of the optoelectronic industry, are widely used in key fields such as industrial processing, medical and health care, and information communication. Market forecasts for lasers directly impact the layout of upstream and downstream supply chains and corporate R&D decisions. Therefore, improving the accuracy of laser market forecasts is a pressing technical challenge.

[0003] In related technologies, laser market forecasting methods mainly rely on historical statistical data and time series models to generate market forecast information for lasers. However, time series models rely on the assumption of stationarity of historical data and can only fit the linear trend of market changes, exhibiting extremely poor responsiveness to non-stationary events such as sudden market fluctuations and industry cyclical declines.

[0004] Therefore, the market forecast information generated by the laser market forecasting methods in related technologies is not very accurate and cannot meet the actual needs of enterprise decision-making and industrial layout. Summary of the Invention

[0005] This application provides a method and apparatus for generating laser market forecast information, which can improve the accuracy of the generated market forecast information.

[0006] In a first aspect, embodiments of this application provide a method for generating laser market forecast information, the method comprising:

[0007] In response to receiving a market forecast request for lasers, the system obtains multi-dimensional market information for lasers, including one or more of the following: industry chain information, industry planning information, industry technology information, industry economic information, market transaction information, and industry popularity information.

[0008] The multidimensional market information corresponding to lasers is preprocessed, and then classified according to time, product type and region to obtain multiple types of text information.

[0009] Feature extraction is performed on multiple types of text information to obtain the index information corresponding to each type of text information. Based on the index information corresponding to each type of text information, a knowledge graph information corresponding to the laser is constructed. The knowledge graph information is used to indicate the causal relationship between multiple types of text information.

[0010] The trained market prediction model is invoked based on the knowledge graph information corresponding to the laser to generate market prediction information corresponding to the laser. The market prediction model is obtained by training the initial market prediction model based on multiple historical market information within a preset time period.

[0011] In one possible design, the multidimensional market information is preprocessed according to the information type, including: if the information type of the multidimensional market information is numerical market information, then outlier removal, missing value imputation, and deduplication are performed on the numerical market information; if the information type of the multidimensional market information is textual market information, then the text information related to lasers is extracted from the textual market information.

[0012] In one possible design, outlier removal, missing value imputation, and deduplication processing are performed on numerical market information, including: outlier removal processing of numerical market information using a preset outlier detection model; and / or missing value imputation processing of numerical market information using a preset interpolation model; and / or deduplication processing of numerical market information using a preset deduplication model.

[0013] In one possible design, laser-related text information is extracted from text-based market information, including: text cleaning of the text-based market information to remove redundant information; and extracting laser-related text information from the text-based market information based on preset keywords and a word frequency inverse document frequency algorithm. The preset keywords include one or more of the following: laser company name, laser technical terms, laser product model, and laser application scenarios.

[0014] In one possible design, feature extraction is performed on multiple types of text information to obtain index information corresponding to each type of text information. This includes: extracting features for each type of text information to obtain text feature vectors corresponding to multiple word segments; determining the mutual information parameter between each text feature vector and market prediction information using a mutual information model; selecting multiple first word segments from multiple word segments whose mutual information parameter is greater than a first preset value, where the mutual information parameter is used to indicate the degree of dependence between the word segment and the market prediction information; determining the importance score of multiple first word segments using a random forest model; and deleting a preset number of target word segments from multiple first word segments based on their importance scores, using these as the index information corresponding to the type of text information.

[0015] In one possible design, a knowledge graph of the laser is constructed based on the index information corresponding to each of multiple types of text information. This includes: constructing a knowledge graph of the laser based on the index information corresponding to each of multiple types of text information and according to a preset association graph. The preset association graph includes at least the relationships between enterprise information, product information, technical information, application information, and customer information.

[0016] In one possible design, the supply chain information includes one or more of the following: component supply information, laser production information, and laser application demand information; the industry technology information includes one or more of the following: patent application information, technological achievement information, and R&D investment information; the industry economic information includes one or more of the following: economic growth rate information, industry GDP share, and per capita disposable income information; the market transaction information includes one or more of the following: laser sales volume information, price information, and regional distribution information; and the industry popularity information includes one or more of the following: industry media reports, social media discussion popularity, and user feedback information.

[0017] In one possible design, the training process of the market prediction model includes: acquiring multiple historical market information sets from the laser within a preset time period, dividing the multiple historical market information sets into a training set, a validation set, and a test set according to a preset ratio and chronological order; training the initial market prediction model using the training set, and determining the prediction error corresponding to the initial market prediction model using the validation set; if the number of training iterations reaches a preset number or the number of consecutive increases in the prediction error reaches a preset number, then the initial market prediction model is determined to be trained successfully, and a trained market prediction model is obtained.

[0018] In one possible design, the method further includes: correcting the market forecast information through a preset rule base; wherein the preset rule base includes the maximum or minimum value corresponding to the market forecast information; the market forecast information includes one or more of the following: market size forecast information, market demand structure forecast information, market price trend forecast information, and regional market performance forecast information.

[0019] Secondly, embodiments of this application provide an apparatus for generating laser market forecast information, the apparatus comprising:

[0020] The acquisition module is used to respond to a market forecast request for lasers and acquire multi-dimensional market information corresponding to lasers. The multi-dimensional market information includes one or more of the following: industry chain information, industry planning information, industry technology information, industry economic information, market transaction information, and industry popularity information.

[0021] The preprocessing module is used to preprocess the multidimensional market information corresponding to the laser, and classify the preprocessed multidimensional market information according to the time dimension, product type dimension and region dimension to obtain multiple types of text information.

[0022] The module is used to extract features from multiple types of text information, obtain the index information corresponding to each type of text information, and construct the knowledge graph information corresponding to the laser based on the index information corresponding to each type of text information. The knowledge graph information is used to indicate the causal relationship between multiple types of text information.

[0023] The market forecasting module is used to call the trained market forecasting model based on the knowledge graph information corresponding to the laser and generate market forecasting information corresponding to the laser. The market forecasting model is obtained by training the initial market forecasting model based on multiple historical market information within a preset time period.

[0024] Thirdly, embodiments of this application provide an electronic device, including: a memory and a processor;

[0025] The memory stores the instructions that the computer executes;

[0026] The processor executes computer execution instructions stored in memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.

[0027] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect.

[0028] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the first aspect and / or various possible implementations of the first aspect.

[0029] The present application provides a method and apparatus for generating laser market forecast information. The method includes: responding to receiving a market forecast request for a laser, acquiring multi-dimensional market information corresponding to the laser, the multi-dimensional market information including one or more of the following: industry chain information, industry planning information, industry technology information, industry economic information, market transaction information, and industry popularity information; preprocessing the multi-dimensional market information corresponding to the laser, and classifying the preprocessed multi-dimensional market information according to time dimension, product type dimension, and region dimension to obtain multiple types of text information; extracting features from the multiple types of text information to obtain index information corresponding to each type of text information; constructing a knowledge graph information corresponding to the laser based on the index information corresponding to each type of text information, the knowledge graph information being used to indicate the causal relationship between the multiple types of text information; and calling a trained market forecast model according to the knowledge graph information corresponding to the laser to generate market forecast information corresponding to the laser. In this embodiment of the disclosure, by preprocessing, extracting features, and constructing knowledge graph information, multidimensional market information is integrated, which solves the prediction error problem caused by poor data quality in traditional market forecasting methods. Furthermore, by constructing knowledge graph information, the sensitivity of the market forecasting model to key market drivers can be improved, ensuring that the market forecasting information can accurately reflect market dynamics and thus improving the accuracy of market forecasting information. Attached Figure Description

[0030] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0031] Figure 1 is a schematic diagram of a scenario for the method of generating laser market forecast information provided in this application;

[0032] Figure 2 is a flowchart illustrating the method for generating laser market forecast information provided in an embodiment of this application;

[0033] Figure 3 is a schematic diagram of the structure of the laser market forecast information generation device provided in an embodiment of this application;

[0034] Figure 4 is a schematic diagram of the structure of the electronic device provided in this application.

[0035] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0036] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0037] Lasers, as core components of the optoelectronic industry, are widely used in key fields such as industrial processing, medical and health care, and information communication. Market forecasts for lasers directly impact the layout of upstream and downstream supply chains and corporate R&D decisions. Therefore, improving the accuracy of laser market forecasts is a pressing technical challenge.

[0038] In related technologies, laser market forecasting methods mainly rely on historical statistical data and time series models to generate market forecast information. However, time series models depend on the stationarity assumption of historical data and can only fit linear market trends, exhibiting extremely poor responsiveness to non-stationary events such as sudden market fluctuations and industry cyclical downturns. Therefore, the market forecast information generated by laser market forecasting methods in related technologies has poor accuracy and cannot meet the actual needs of corporate decision-making and industrial layout.

[0039] With the development of laser technology, higher requirements are placed on the accuracy, timeliness and depth of laser market forecasting. There is an urgent need for an efficient forecasting method that can integrate multi-source information, capture complex correlations and adapt to dynamic changes.

[0040] To address the aforementioned technical problems, the inventors propose the following technical concept: This application acquires multiple factors such as upstream and downstream of the industrial chain, technological iteration, and macroeconomic conditions, and combines structured information such as sales volume and price with unstructured information such as patent texts and industry news. By constructing a knowledge graph, it integrates industry-related information and mines the implicit relationships between "enterprises-products-technology-applications-customers" to support complex market reasoning needs. This proposes a laser market forecasting method that integrates multimodal information, achieves dynamic updates, and possesses deep causal reasoning capabilities.

[0041] The specific steps may include: in response to receiving a market forecast request for a laser, obtaining multi-dimensional market information corresponding to the laser, including one or more of the following: industry chain information, industry technology information, industry economic information, market transaction information, and industry popularity information; preprocessing the multi-dimensional market information corresponding to the laser, and classifying the preprocessed multi-dimensional market information according to time dimension, product type dimension, and region dimension to obtain multiple types of text information; extracting features from the multiple types of text information to obtain the index information corresponding to each type of text information; constructing a knowledge graph information corresponding to the laser based on the index information corresponding to each type of text information, the knowledge graph information being used to indicate the causal relationship between the multiple types of text information; and calling a trained market forecast model based on the knowledge graph information corresponding to the laser to generate market forecast information corresponding to the laser.

[0042] In this embodiment of the disclosure, by preprocessing, extracting features, and constructing knowledge graph information, multidimensional market information is integrated, which solves the prediction error problem caused by poor data quality in traditional market forecasting methods. Furthermore, by constructing knowledge graph information, the sensitivity of the market forecasting model to key market drivers can be improved, ensuring that the market forecasting information can accurately reflect market dynamics and thus improving the accuracy of market forecasting information.

[0043] The collection, storage, use, processing, transmission, provision, and disclosure of financial data or user data involved in the technical solution of this application all comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0044] It should be noted that in the embodiments of this application, certain software, components, models and other existing solutions in the industry may be mentioned. These should be regarded as exemplary and are only intended to illustrate the feasibility of implementing the technical solution of this application. However, it does not mean that the applicant has used or necessarily used the solution.

[0045] Figure 1 is a schematic diagram of a scenario for the method of generating laser market forecast information provided in this application. As shown in Figure 1, the scenario includes: server 101 and terminal 102.

[0046] In the specific implementation process, server 101 and terminal 102 can be implemented using a cluster of one or more servers with more powerful processing capabilities and higher security. Where possible, computers or laptops with strong computing power can also be used as alternatives.

[0047] The connection between server 101 and terminal 102 can be wired or wireless. A user can send a market forecast request for the laser to server 101 via terminal 102. Server 101 receives the market forecast request and generates market forecast information corresponding to the laser using the laser market forecast information generation method provided in this embodiment. Server 101 returns the generated market forecast information to terminal 102, which then displays the market forecast information.

[0048] It is understood that the scenarios illustrated in the embodiments of this application do not constitute a specific limitation on the method for generating laser market forecast information. In other feasible embodiments of this application, the above scenarios may include more or fewer components than illustrated, or combine some components, or split some components, or arrange different components, which can be determined according to the actual application scenario and is not limited here. The scenario shown in Figure 1 can be implemented by hardware, software, or a combination of software and hardware.

[0049] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0050] Figure 2 is a flowchart illustrating the method for generating laser market forecast information provided in an embodiment of this application. The executing entity in this embodiment can be the server 101 in Figure 1, or a computer and / or mobile phone, etc., and this embodiment does not impose any particular limitation. As shown in Figure 2, the method includes:

[0051] S201. In response to receiving a market forecast request for lasers, obtain multi-dimensional market information corresponding to lasers. The multi-dimensional market information includes one or more of the following: industry chain information, industry planning information, industry technology information, industry economic information, market transaction information, and industry popularity information.

[0052] In this embodiment of the application, the market forecast request for lasers can be a market forecast request corresponding to any market information. This market information can include market size, market demand structure, market price trends, regional market performance, etc. For example, the market forecast request can be a forecast request for predicting the laser market, a forecast request for predicting the demand share of various types of lasers, or a forecast request for predicting the price fluctuation range of a certain product type.

[0053] The multidimensional market information can be any market information related to lasers. Optionally, the industry chain information includes one or more of the following: component supply information, laser production information, and laser application demand information; industry technology information includes one or more of the following: patent application information, technological achievement information, and R&D investment information; industry economic information includes one or more of the following: economic growth rate information, industry GDP share, and per capita disposable income information; market transaction information includes one or more of the following: laser sales volume information, price information, and regional distribution information; and industry popularity information includes one or more of the following: industry media reports, social media discussion popularity, and user feedback information.

[0054] For example, industry technology information includes the number of laser-related patent applications and the amount of investment in technology research and development. Industry economic information includes the industry's share of GDP and the manufacturing PMI (Purchasing Managers Index). Market transaction information includes monthly sales and prices of several mainstream products such as semiconductor lasers and solid-state lasers. Industry trending information includes relevant reports from industry media and texts related to the topic of "laser technology" on social media.

[0055] In some embodiments, a data acquisition module can be used to obtain multi-dimensional market information corresponding to lasers. This data acquisition module includes a multi-source data interface for acquiring multi-dimensional market information from third-party platforms such as industry databases. For example, the multi-source data interface of the data acquisition module can be used to connect to industry databases and patent databases. For instance, patent information related to lasers can be obtained through a patent API interface. This patent information may include patent text information, patent classification number information, and patent status information. For example, market size, shipment volume, and other data can be obtained from public databases.

[0056] S202. Preprocess the multi-dimensional market information corresponding to the laser, and classify the preprocessed multi-dimensional market information according to the time dimension, product type dimension and region dimension to obtain multiple types of text information.

[0057] In some embodiments, the collected raw data is cleaned, integrated, and feature-engineered to provide high-quality data for model input. Correspondingly, the multidimensional market information corresponding to the laser is preprocessed, including: if the information type of the multidimensional market information is numerical market information, outlier removal, missing value imputation, and deduplication are performed on the numerical market information; if the information type of the multidimensional market information is textual market information, the textual information related to the laser is extracted from the textual market information.

[0058] Numerical market information can be structured data such as "component supply rate" and "capacity utilization rate." Textual market information can be unstructured data such as patent texts and industry media reports.

[0059] Optionally, outlier removal, missing value imputation, and deduplication processing are performed on numerical market information, including: outlier removal processing of numerical market information using a preset outlier detection model; and / or missing value imputation processing of numerical market information using a preset interpolation model; and / or deduplication processing of numerical market information using a preset deduplication model.

[0060] In this application, the outlier detection model, interpolation model, and deduplication model are not specifically limited. For example, the outlier detection model can be a Z-score algorithm model or an IQR (interquartile range) algorithm model. The Z-score algorithm model identifies outliers by calculating the multiple of deviation between a data point and the mean. The IQR algorithm model calculates the upper quartile (Q3) and lower quartile (Q1) of a data point to obtain the interquartile range IQR = Q3 - Q1, and identifies data points less than Q1 - 1.5 × IQR or greater than Q3 + 1.5 × IQR as outliers. For example, the interpolation model can be a mean interpolation model, a median interpolation model, or a mode interpolation model, etc. For example, the deduplication model can be a hash deduplication model or a rule-matching deduplication model, etc. The hash deduplication model generates a unique hash value for each data point; if two data points have the same hash value, they are considered duplicate data.

[0061] In some embodiments, natural language processing techniques are used to perform feature transformation on unstructured data (such as text-based market information) to extract text features. Optionally, extracting laser-related text information from text-based market information includes: text cleaning of the text-based market information to remove redundant information; and extracting laser-related text information from the text-based market information based on preset keywords and a word frequency inverse document frequency algorithm. The preset keywords include one or more of the following: laser company name, laser technical terms, laser product model, and laser application scenarios.

[0062] For example, text-based market information can be patent text. By performing text cleaning on the patent text, redundant information is removed, and the technology-related text information is retained. The redundant information can be preset format information, clause information, etc.

[0063] The inverse document frequency (IVF) algorithm is used to measure the importance of keywords in text. Optionally, based on preset keywords and the IVF algorithm, text information related to lasers in text-based market information is extracted, including: determining the importance parameter of each preset keyword in the text-based market information based on the preset keywords and the IVF algorithm; selecting multiple target keywords from multiple preset keywords whose importance parameter is greater than a preset importance parameter threshold, and identifying these multiple target keywords as the laser-related text information in the text-based market information.

[0064] In this embodiment, the preprocessed multidimensional market information can be classified according to one or more dimensions, including time, product type, and region, to obtain multiple types of text information. The time dimension can be month, quarter, year, etc. The region dimension can include region A, region B, etc.

[0065] For example, text information can be categorized and integrated according to time (e.g., monthly, quarterly), product type (e.g., solid-state laser, gas laser, semiconductor laser, etc.), and region to obtain multiple types of text information.

[0066] S203. Perform feature extraction on multiple types of text information to obtain the index information corresponding to each type of text information. Based on the index information corresponding to each type of text information, construct the knowledge graph information corresponding to the laser. The knowledge graph information is used to indicate the causal relationship between multiple types of text information.

[0067] In some embodiments, feature extraction is performed on multiple types of text information to obtain index information corresponding to each type of text information, including: extracting features for each type of text information to obtain text feature vectors corresponding to multiple word segments; determining the mutual information parameter between each text feature vector and market prediction information through a mutual information model; selecting multiple first word segments from multiple word segments whose mutual information parameter is greater than a first preset value, wherein the mutual information parameter is used to indicate the degree of dependence between the word segment and the market prediction information; determining the importance score of multiple first word segments through a random forest model; and deleting a preset number of target word segments from multiple first word segments based on the importance score of the multiple first word segments, which are used as index information corresponding to the type of text information.

[0068] In this embodiment, the mutual information parameter is an indicator that measures the degree of dependence between two random variables, and its value ranges from [0, +∞). A higher value for the mutual information parameter indicates a stronger dependence between the text features and the prediction target. The mutual information parameter can capture both linear and non-linear dependencies, thereby improving the data quality for generating market prediction information.

[0069] The market forecast information may include one or more of the following: market size forecast information, market demand structure forecast information, market price trend forecast information, and regional market performance forecast information. For example, if the market forecast information includes market size forecast information, the degree of dependence between the segmented word and the market forecast information can be determined by the mutual information parameter between the text feature vector corresponding to the segmented word and the market forecast information.

[0070] In a random forest model, multiple decision trees are ensembled. Feature importance is measured by the reduction in the Gini coefficient or "information gain" when the feature splits across all decision trees. The higher the frequency with which a feature is selected in tree splits, the higher its importance score. This importance score represents the feature's contribution to the model's predictions.

[0071] In some embodiments, knowledge graph information corresponding to the laser is constructed based on the index information corresponding to each of multiple types of text information, including: constructing knowledge graph information corresponding to the laser according to a preset association graph based on the index information corresponding to each of multiple types of text information, wherein the preset association graph includes at least the association relationships between enterprise information, product information, technical information, application information and customer information.

[0072] Optionally, a dynamically updated knowledge graph can be formed by automatically extracting the relationships between "enterprises, products, technologies, applications, and customers" based on a preset association graph. This knowledge graph can use a GNN (Graph Neural Network) to generate entity relationship vectors, and its timeliness can be maintained through real-time data acquisition and event extraction. For example, when a "breakthrough in semiconductor laser technology" event is detected, an automatic correlation analysis of the decrease in fiber laser costs and the increase in demand for new energy sources can be triggered.

[0073] S204. Based on the knowledge graph information corresponding to the laser, call the trained market prediction model to generate market prediction information corresponding to the laser. The market prediction model is obtained by training the initial market prediction model based on multiple historical market information within a preset time period.

[0074] In this embodiment, the type of market forecasting model is not specifically limited. The market forecasting model can be a Transformer model, or a hybrid model combining machine learning and AI (Artificial Intelligence) models. The input to this market forecasting model is the knowledge graph information corresponding to the laser, and the output is the market forecast information corresponding to the laser.

[0075] Knowledge graph information is used to indicate the relationships between multiple types of text information in order to improve the accuracy of market forecasting information.

[0076] For example, the market forecast information for lasers includes multi-dimensional forecast results such as laser market size information (e.g., total market value in the next 1-3 years), demand structure information (e.g., demand share of various laser products and demand distribution in downstream application fields), price trend information (e.g., price fluctuation range of mainstream products) and regional market performance information (e.g., market growth rate in each region).

[0077] In some embodiments, industry experience can be incorporated to refine the market forecast information output by the model. Optionally, the market forecast information can be refined using a pre-defined rule base to ensure its reasonableness.

[0078] In one possible design, the method further includes: correcting the market forecast information through a preset rule base; wherein the preset rule base includes the maximum or minimum value corresponding to the market forecast information; the market forecast information includes one or more of the following: market size forecast information, market demand structure forecast information, market price trend forecast information, and regional market performance forecast information.

[0079] For example, the maximum or minimum demand share of laser product A. If the demand share of laser product A in the market forecast information is greater than the maximum demand share of laser product A in the rule base, then the market forecast information will be corrected to the maximum demand share of laser product A in the rule base.

[0080] Optionally, the market forecast information corresponding to the laser can be displayed in multiple ways. Optionally, it can be displayed through data reports, visualization charts (such as line charts, bar charts, and heat maps), dynamic reports, etc., to facilitate users' intuitive access to and use of market forecast information. Furthermore, the market forecast information can be converted into files in preset formats, thus facilitating its export. These preset formats can be Excel, PDF, etc.

[0081] The method for generating laser market forecast information provided in this application embodiment is as follows: In response to receiving a market forecast request for a laser, multi-dimensional market information corresponding to the laser is obtained. The multi-dimensional market information includes one or more of the following: industry chain information, industry planning information, industry technology information, industry economic information, market transaction information, and industry popularity information. The multi-dimensional market information corresponding to the laser is preprocessed and classified according to time dimension, product type dimension, and region dimension to obtain multiple types of text information. Features are extracted from the multiple types of text information to obtain index information corresponding to each type of text information. Based on the index information corresponding to each type of text information, a knowledge graph information corresponding to the laser is constructed. The knowledge graph information is used to indicate the causal relationship between the multiple types of text information. The trained market forecast model is called according to the knowledge graph information corresponding to the laser to generate the market forecast information corresponding to the laser. The market forecast model is obtained by training an initial market forecast model based on multiple historical market information within a preset time period. In this embodiment of the disclosure, by preprocessing, extracting features, and constructing knowledge graph information, multidimensional market information is integrated, which solves the prediction error problem caused by poor data quality in traditional market forecasting methods. Furthermore, by constructing knowledge graph information, the sensitivity of the market forecasting model to key market drivers can be improved, ensuring that the market forecasting information can accurately reflect market dynamics and thus improving the accuracy of market forecasting information.

[0082] The training process of the market forecasting model is described below: In some embodiments, the training process of the market forecasting model includes the following steps (1) to (3):

[0083] (1) Obtain multiple historical market information of the laser within a preset time period, and divide the multiple historical market information into training set, validation set and test set according to preset ratio and time sequence.

[0084] In this embodiment, the values ​​of the preset duration and preset ratio are not specifically limited. Optionally, multiple historical market information sets are divided into a training set, a validation set, and a test set in a 7:2:1 ratio. The training set is used for fitting model parameters; the validation set is used for adjusting model hyperparameters and monitoring overfitting; and the test set is used for final model performance evaluation.

[0085] For example, historical market information from January 2018 to December 2024 is divided into a 7:2:1 set: training set (January 2018 to December 2022), validation set (January 2023 to June 2024), and test set (July 2024 to December 2024).

[0086] (2) The initial market prediction model is trained using the training set, and the prediction error corresponding to the initial market prediction model is determined using the validation set.

[0087] In this embodiment of the application, when training the initial market prediction model using the training set, steps S201 to S203 can be used to determine the knowledge graph information corresponding to each historical market information in the training set, and input the knowledge graph information into the initial market prediction model for training.

[0088] (3) If the number of training iterations reaches the preset number of iterations or the number of consecutive increases in the prediction error reaches the preset number of iterations, then the initial market prediction model training is completed and a well-trained market prediction model is obtained.

[0089] In this embodiment, the value of the preset number of iterations is not specifically limited. Optionally, the preset number of iterations can be 100. Optionally, the feature data in the training set and the corresponding prediction target (such as the market size of lasers in the future quarter, the sales share of a certain type of product, etc.) are input into the market prediction model, initial hyperparameters (such as learning rate, number of iterations, batch size) are set, and the gradient descent algorithm is used to train the model. During the training process, the prediction error of the model (such as mean squared error or mean absolute percentage error) is monitored in real time through the validation set. When the error of the validation set increases continuously for multiple rounds, it is determined that the initial market prediction model training is complete, and the trained market prediction model is obtained.

[0090] Here, batch size refers to the number of samples processed before each parameter update. For example, the initial learning rate is set to 0.001, the batch size to 32, and the number of iterations to 100 epochs, using the Adam algorithm for training. During training, an early stopping strategy is triggered when the MAPE error on the validation set increases for three consecutive epochs, ultimately stopping training at the 68th epoch.

[0091] In some embodiments, the method further includes: optimizing the model hyperparameters using methods such as grid search and Bayesian optimization to determine the optimal combination of hyperparameters. To address the problem of uneven data distribution during model training (such as limited data for certain laser products), data augmentation techniques (such as time-series data resampling and synthesizing minority class samples) are employed for optimization.

[0092] For example, the learning rate of the market forecasting model is determined to be 0.0008 and the batch size is 64 through Bayesian optimization.

[0093] It should be noted that the embodiments of this application can also combine Bayesian networks and PC algorithms. The PC algorithm extracts causal graphs (knowledge graph information) from historical data to identify causal relationships in market changes (such as the causal relationship between breakthroughs in fiber laser technology and the growth in demand for new energy vehicles). Bayesian networks, by quantifying the strength of causal relationships and incorporating causal inference results into the market prediction model through an attention mechanism, further improve the accuracy of market prediction information generated by the market prediction model.

[0094] Figure 3 is a schematic diagram of the structure of the laser market forecast information generation device provided in an embodiment of this application. As shown in Figure 3, the laser market forecast information generation device includes:

[0095] The acquisition module 301 is used to respond to receiving a market forecast request for lasers and acquire multi-dimensional market information corresponding to lasers. The multi-dimensional market information includes one or more of the following: industry chain information, industry planning information, industry technology information, industry economic information, market transaction information, and industry popularity information.

[0096] The preprocessing module 302 is used to preprocess the multi-dimensional market information corresponding to the laser, and classify the preprocessed multi-dimensional market information according to the time dimension, product type dimension and region dimension to obtain multiple types of text information.

[0097] The construction module 303 is used to extract features from multiple types of text information, obtain the index information corresponding to each type of text information, and construct the knowledge graph information corresponding to the laser based on the index information corresponding to each type of text information. The knowledge graph information is used to indicate the causal relationship between multiple types of text information.

[0098] The market forecasting module 304 is used to call the trained market forecasting model based on the knowledge graph information corresponding to the laser to generate market forecasting information corresponding to the laser; wherein the market forecasting model is obtained by training the initial market forecasting model based on multiple historical market information within a preset time period.

[0099] In one possible design, the preprocessing module 302 preprocesses the multidimensional market information according to the information type, including: if the information type of the multidimensional market information is numerical market information, then outlier removal, missing value filling and deduplication are performed on the numerical market information; if the information type of the multidimensional market information is text market information, then the text information related to the laser is extracted from the text market information.

[0100] In one possible design, the preprocessing module 302 performs outlier removal, missing value imputation, and deduplication on the numerical market information, including: performing outlier removal on the numerical market information using a preset outlier detection model; and / or performing missing value imputation on the numerical market information using a preset interpolation model; and / or performing deduplication on the numerical market information using a preset deduplication model.

[0101] In one possible design, the preprocessing module 302 extracts laser-related text information from text-based market information, including: cleaning the text-based market information and removing redundant information; and extracting laser-related text information from the text-based market information based on preset keywords and a word frequency inverse document frequency algorithm. The preset keywords include one or more of the following: laser company name, laser technical terms, laser product model, and laser application scenarios.

[0102] In one possible design, the construction module 303 performs feature extraction on multiple types of text information to obtain index information corresponding to each type of text information. This includes: extracting features for each type of text information to obtain text feature vectors corresponding to multiple word segments; determining the mutual information parameter between each text feature vector and market prediction information through a mutual information model; selecting multiple first word segments from multiple word segments whose mutual information parameter is greater than a first preset value, where the mutual information parameter is used to indicate the degree of dependence between the word segment and the market prediction information; determining the importance score of multiple first word segments through a random forest model; and deleting a preset number of target word segments from multiple first word segments based on their importance scores, using these as index information corresponding to the type of text information.

[0103] In one possible design, the construction module 303 constructs the knowledge graph information corresponding to the laser based on the index information corresponding to each of the multiple types of text information, including: constructing the knowledge graph information corresponding to the laser according to a preset association graph based on the index information corresponding to each of the multiple types of text information, wherein the preset association graph includes at least the association relationships between enterprise information, product information, technical information, application information and customer information.

[0104] In one possible design, the supply chain information includes one or more of the following: component supply information, laser production information, and laser application demand information; the industry technology information includes one or more of the following: patent application information, technological achievement information, and R&D investment information; the industry economic information includes one or more of the following: economic growth rate information, industry GDP share, and per capita disposable income information; the market transaction information includes one or more of the following: laser sales volume information, price information, and regional distribution information; and the industry popularity information includes one or more of the following: industry media reports, social media discussion popularity, and user feedback information.

[0105] In one possible design, the training process of the market prediction model includes: acquiring multiple historical market information sets from the laser within a preset time period, dividing the multiple historical market information sets into a training set, a validation set, and a test set according to a preset ratio and chronological order; training the initial market prediction model using the training set, and determining the prediction error corresponding to the initial market prediction model using the validation set; if the number of training iterations reaches a preset number or the number of consecutive increases in the prediction error reaches a preset number, then the initial market prediction model is determined to be trained successfully, and a trained market prediction model is obtained.

[0106] In one possible design, the device further includes: a correction module; the correction module is used to correct the market forecast information through a preset rule base; wherein the preset rule base includes the maximum or minimum value corresponding to the market forecast information; the market forecast information includes one or more of market size forecast information, market demand structure forecast information, market price trend forecast information, and regional market performance forecast information.

[0107] The laser market forecasting information generation device provided in this application embodiment solves the prediction error problem caused by poor data quality in traditional market forecasting methods by integrating multi-dimensional market information through preprocessing, feature extraction, and knowledge graph construction. Furthermore, the constructed knowledge graph information can improve the sensitivity of the market forecasting model to key market drivers, ensuring that the market forecasting information can accurately reflect market dynamics, thus improving the accuracy of market forecasting information.

[0108] The laser market forecast information generation device provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0109] Figure 4 is a schematic diagram of the structure of the electronic device provided in this application. As shown in Figure 4, the electronic device 40 provided in this embodiment includes at least one processor 401 and a memory 402. Optionally, the electronic device 40 further includes a communication component 403. The processor 401, the memory 402, and the communication component 403 are connected via a bus.

[0110] In a specific implementation, at least one processor 401 executes computer execution instructions stored in memory 402, causing at least one processor 401 to perform the above-described method.

[0111] The specific implementation process of processor 401 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0112] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0113] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0114] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0115] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0116] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0117] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0118] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0119] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0120] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0121] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0122] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0123] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0124] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A method for generating laser market forecast information, characterized in that, include: In response to a market forecast request for a laser, the system acquires multi-dimensional market information corresponding to the laser, including one or more of the following: industry chain information, industry planning information, industry technology information, industry economic information, market transaction information, and industry popularity information. The system preprocesses the multi-dimensional market information corresponding to the laser and classifies it according to time, product type, and region dimensions to obtain multiple types of text information. Feature extraction is performed on the multiple types of text information to obtain index information corresponding to each type. Based on the index information corresponding to each type of text information, a knowledge graph information corresponding to the laser is constructed, which indicates the causal relationships between the multiple types of text information. A trained market forecast model is invoked based on the knowledge graph information corresponding to the laser to generate market forecast information corresponding to the laser, wherein the market forecast model is trained on an initial market forecast model based on multiple historical market information within a preset time period.

2. The method according to claim 1, characterized in that, The preprocessing of the multidimensional market information corresponding to the laser includes: if the information type of the multidimensional market information is numerical market information, then outlier removal, missing value filling and deduplication processing are performed on the numerical market information; if the information type of the multidimensional market information is textual market information, then the textual market information related to the laser is extracted.

3. The method according to claim 2, characterized in that, The outlier removal, missing value imputation, and deduplication processing of the numerical market information includes: performing outlier removal processing on the numerical market information using a preset outlier detection model; and / or performing missing value imputation processing on the numerical market information using a preset interpolation model; and / or performing deduplication processing on the numerical market information using a preset deduplication model.

4. The method according to claim 2, characterized in that, The step of extracting text information related to the laser from the text-based market information includes: cleaning the text-based market information and deleting redundant information from it; and extracting text information related to the laser from the text-based market information based on preset keywords and a word frequency inverse document frequency algorithm. The preset keywords include one or more of the following: laser company name, laser technical terms, laser product model, and laser application scenario.

5. The method according to claim 1, characterized in that, The step of extracting features from the multiple types of text information to obtain index information corresponding to each type of text information includes: extracting features from each type of text information to obtain text feature vectors corresponding to multiple word segments; determining the mutual information parameter between each text feature vector and market prediction information using a mutual information model; selecting multiple first word segments from the multiple word segments whose mutual information parameter is greater than a first preset value, wherein the mutual information parameter is used to indicate the degree of dependence between the word segment and the market prediction information; determining the importance score of the multiple first word segments using a random forest model; and deleting a preset number of target word segments from the multiple first word segments based on the importance score of the multiple first word segments, which are then used as the index information corresponding to the text information of the aforementioned type.

6. The method according to claim 1, characterized in that, The step of constructing the knowledge graph information corresponding to the laser based on the index information corresponding to each of the multiple types of text information includes: constructing the knowledge graph information corresponding to the laser based on the index information corresponding to each of the multiple types of text information according to a preset association graph, wherein the preset association graph includes at least the association relationships between enterprise information, product information, technical information, application information and customer information.

7. The method according to any one of claims 1-6, characterized in that, The industry chain information includes one or more of the following: component supply information, laser production information, and laser application demand information; the industry technology information includes one or more of the following: patent application information, technological achievement information, and R&D investment information; the industry economic information includes one or more of the following: economic growth rate information, industry GDP share, and per capita disposable income information; the market transaction information includes one or more of the following: laser sales volume information, price information, and regional distribution information; and the industry popularity information includes one or more of the following: industry media reports, social media discussion popularity, and user feedback information.

8. The method according to claim 1, characterized in that, The training process of the market prediction model includes: acquiring multiple historical market information from the laser within a preset time period, dividing the multiple historical market information into a training set, a validation set, and a test set according to a preset ratio and chronological order; training the initial market prediction model using the training set, and determining the prediction error corresponding to the initial market prediction model using the validation set; if the number of training iterations reaches a preset number or the number of consecutive increases in the prediction error reaches a preset number, then the initial market prediction model is determined to be trained successfully, and a trained market prediction model is obtained.

9. The method according to claim 1, characterized in that, The method further includes: correcting the market forecast information through a preset rule base; wherein the preset rule base includes the maximum or minimum value corresponding to the market forecast information; the market forecast information includes one or more of market size forecast information, market demand structure forecast information, market price trend forecast information, and regional market performance forecast information.

10. A device for generating laser market forecast information, characterized in that, include: The acquisition module is used to respond to a market forecast request for a laser by acquiring multi-dimensional market information corresponding to the laser, including one or more of the following: industry chain information, industry planning information, industry technology information, industry economic information, market transaction information, and industry popularity information. The preprocessing module is used to preprocess the multi-dimensional market information corresponding to the laser and classify it according to time, product type, and region dimensions to obtain multiple types of text information. The construction module is used to extract features from the multiple types of text information to obtain index information corresponding to each type of text information, and to construct a knowledge graph information corresponding to the laser based on the index information corresponding to each type of text information. The knowledge graph information is used to indicate the causal relationships between the multiple types of text information. The market forecasting module is used to call a trained market forecasting model based on the knowledge graph information corresponding to the laser to generate market forecast information corresponding to the laser, wherein the market forecasting model is trained on an initial market forecasting model based on multiple historical market information within a preset time period.

Images